Continuation Method for Feedback Delay Network Modal Decomposition
Bai B, Schlecht S (2026)
Publication Language: English
Publication Type: Conference contribution, Conference Contribution
Publication year: 2026
Pages Range: 16002 - 16006
Conference Proceedings Title: IEEE ICASSP 2026 Proceedings
Event location: Barcelona, Spain
URI: https://ieeexplore.ieee.org/document/11460431
DOI: 10.1109/ICASSP55912.2026.11460431
Abstract
Feedback Delay Networks (FDNs) modal decomposition requires solving large polynomial eigenvalue problems, which is costly when the feedback matrix varies. We propose a continuation approach that tracks poles along matrix homotopies using eigenderivatives and a predictor–corrector scheme. Linear and exponential paths are compared, with phase-only updates preserving lossless cases. Experiments on moderately sized FDNs show smooth trajectories, minor pole loss, and reasonable computational cost compared to the Ehrlich–Aberth method, supporting both modal and gradient-based analysis.
Authors with CRIS profile
How to cite
APA:
Bai, B., & Schlecht, S. (2026). Continuation Method for Feedback Delay Network Modal Decomposition. In IEEE ICASSP 2026 Proceedings (pp. 16002 - 16006). Barcelona, Spain, ES.
MLA:
Bai, Baoqi, and Sebastian Schlecht. "Continuation Method for Feedback Delay Network Modal Decomposition." Proceedings of the 2026 IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP), Barcelona, Spain 2026. 16002 - 16006.
BibTeX: Download